|Summary||Big data approach to predictive maintenance|
|Supervisor||Stefan Byttner, Cristofer Englund|
This project is based on a need for improved analysis functions to enable predictive maintenance of transport system infrastructure. The purpose is to use historical data to build prediction models in order to improve quality and reduce cost for maintenance of transport system infrastructure.
Data is available through a cloud service, Infracontrol Online, hosted by Infracontrol AB in Mölndal and is used by a number of cities in Sweden. The data consists of reports that are generated as long as errors are reported. The typical errors that are reported are broken street lights, broken traffic lights, slippery road conditions, problem with snow and problem with vegetation.
Other areas where this type of methodology is successfully used is e.g. predictive policing. Researchers have found patterns in historical data of police reports and used them as a base for more intelligent police patroling.
The overall project is divided into the following tasks:
1. Identification of what services have the highest potential to improve service.
2. Extract data to be used to develop algorithms and methodology.
3. Model development of predictive analysis model
4. Visualization of analysis results
5. Model evaluation during long term use of application.
The thesis work will focus on the second and third tasks.